Reservoir characterization using dynamic welltest/production and microseismic data

The goal of this research is to integrate dynamic welltest/production and microseismic data to obtain accurate reservoir characterization, which is necessary for early stage reservoir development planning. The early time dynamic data may include a few days of production and subsequent shut-in pressure data. These dynamic data can be integrated to reduce the uncertainties of the reservoir models. However, it is not possible to resolve the layer reservoir rock properties with these dynamic data alone. One possible solution is to collect production logging data, where layer production rates can be inferred. Here, we also explore another possible solution, namely integration of microseismic data which are available during perforation. Microseismic technology has gained popularity recently with the development of multi-stage hydraulic fracturing in shale gas reservoirs, but the application of microseismic technology is mainly limited to fracture characterization. In this research, we explore the application of microseismic data to reservoir porosity and permeability field characterization which would be beneficial in both conventional reservoir and unconventional shale gas reservoirs.;The transient pressure data can resolve the thickness-weighted average permeability in a layered reservoir but are sensitive to the log-permeability of the high porosity, high permeability layers while the microseismic data are more sensitive to the porosities of the low porosity (high velocity) layers. Therefore, these two types of data are complementary and the integration of both types of data can improve the accuracy of the reservoir characterization.;The forward model that is used to calculate the first arrival times is the finite-difference solution of the Eikonal equation. The forward model that is used to predict production data is the commercial simulator ECLIPSE 100. We use the ensemble Kalman filter (EnKF) to assimilate the data. The EnKF does not require computing the gradient of an objective function, and it can be coupled with any forward model easily. In the procedure for integrating production/pressure data and microseismic data considered here, the static geological/geophysical data are assumed to be encapsulated in a multivariate probability density function characterized by a prior mean and covariance for the joint distribution of the porosity and permeability fields. The method is tested with synthetic reservoir models. Excellent data matches are obtained with EnKF and the observed data fall within the uncertainty bounds of the ensemble data predictions.;In the microseismic event location inversion study, we first present an efficient gradient-based method. A novel method is devised to obtain the gradient of the first arrival times to the event location parameters in addition to the first arrival times in one forward model run. The method is applied to a simple one-stage of hydraulic fracture and obtained good event location parameter estimation. Since the arrival time is least sensitive to the event coordinate in the axis where source and receiver are closest, estimation of the coordinate in this direction is least accurate among all coordinates. EnKF is applied to this same event location inversion case and similar results are obtained. However, the ensemble-base method is advantageous in capturing uncertainties in velocity structure.